Severe spermatogenic failure (SF) affects nearly 1% of all men, accounting for approximately 20% of all cases of male infertility, and manifests as either scarcity (oligozoospermia) or absence (azoospermia) of sperm in ejaculate. Physicians are severely limited in their ability to offer useful direction for SF patients pertaining to the chances for successful treatment using assisted reproductive techniques (ART) and potential negative consequences in their ART-derived offspring because more than half of SF cases currently have no identifiable cause. Nonetheless, clinical members of our group have used genetic testing routinely in the management of male infertility for nearly a decade, as several Y chromosome microdeletions are known to be diagnostic of spermatogenic failure, and are prognostic for the success of surgical sperm retrieval. Over 1,000 genes are required for spermatogenesis in mammals, and over 200 mouse knockouts manifest SF; we hypothesize that over 20% of idiopathic SF cases are due to single-gene or oligogenic defects. We have recently published an array-based pilot study demonstrating that numerous genetic causes of SF remain to be discovered, and that recurrent deletions of the gene DMRT1 are a frequent, novel genetic cause of SF. In this proposal we describe experiments that will dramatically expand upon these findings to identify novel genetic causes of SF. We will perform whole exome sequencing of 1000 idiopathic azoospermic men, and 1000 control men with normal sperm concentrations. With these data we will produce an integrated map of genetic variants, using special computational and experimental reagents to enhance our analysis of the repeat-rich Y chromosome. We will perform a genomewide case-control analysis using untargeted variant and gene-based testing, as well as testing numerous biological hypotheses by aggregating variants into pathways or genomic annotations. We propose to use an efficient two-stage design to replicate the top 1% of associated variants and top 5% of associated genes. We have designed the stage I studies for maximum sensitivity to 3 distinct forms of SF mutations. In stage I(a) we re-use the control data generated as part of aim 1 to replicate common risk factors of moderate effect using a quantitative-trait analysis of sperm count. In stage I(b) we use an extreme phenotype sampling strategy to sample new patients from the top and bottom 10% of the sperm count distribution, maximizing our power to detect extremely rare monogenic/Mendelian variants of large effect. In stage I(c) we use existing genetic data from women with primary ovarian insufficiency (POI) to test for genes/genetic variants that modulate gonadal function in women as well as men. Variants of interest that are replicated in at least one of I(a,b,c) will then be taken forward for validation in a Chinese cohort of 3,000 azoospermic cases and 3,000 controls. In the short term, our findings will improve the ability of clinicians to use genetic data to counsel patients and improve patient care. In the long term, characterization of the genetic basis of spermatogenic failure could pave the way for the use of gene therapy strategies in azoospermic men.